Publication Type
Master Thesis
Version
publishedVersion
Publication Date
8-2026
Abstract
This paper develops a framework to bound policy relevant treatment effects in the presence of endogenous sample selection. Since the always-observed subpopulation cannot be directly identified from the data, we first derive moment equalities based on a class of IV-like estimands for this subpopulation and then apply a trimming procedure to transform these equalities into moment inequalities. Under two sets of assumptions on the sample selection mechanism, we establish identification results for these inequalities. The inequalities impose linear restrictions on the parameters of interest, which can be further tightened by incorporating shape restrictions. We formulate the identification problem as a linear programming problem and derive informative upper and lower bounds for a wide range of parameters, including ex-trapolated local average treatment effects (LATEs) and policy relevant treatment effects (PRTEs). We illustrate the empirical relevance of our framework using the
health insurance data from Deb et al. (2006).
Keywords
Sample selection, Partial identification, Instrumental Variables, Extrapolation
Degree Awarded
Master of Philosophy in Econ
Discipline
Econometrics
Supervisor(s)
ZHANG, Yichong
First Page
1
Last Page
64
Publisher
Singapore Management University
City or Country
Singapore
Citation
ZHANG, Pengyang.
Policy relevant treatment effects under sample selection. (2026). 1-64.
Available at: https://ink.library.smu.edu.sg/etd_coll/920
Copyright Owner and License
Author
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.